CN111625954B - Parallel optimization method and system for rainfall landslide model TRIGRS - Google Patents
Parallel optimization method and system for rainfall landslide model TRIGRS Download PDFInfo
- Publication number
- CN111625954B CN111625954B CN202010442518.6A CN202010442518A CN111625954B CN 111625954 B CN111625954 B CN 111625954B CN 202010442518 A CN202010442518 A CN 202010442518A CN 111625954 B CN111625954 B CN 111625954B
- Authority
- CN
- China
- Prior art keywords
- rainfall
- parallel
- data
- trigrs
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 151
- 238000005457 optimization Methods 0.000 title claims abstract description 55
- 230000008569 process Effects 0.000 claims abstract description 115
- 238000004364 calculation method Methods 0.000 claims abstract description 52
- 230000035515 penetration Effects 0.000 claims abstract description 19
- 230000000694 effects Effects 0.000 claims abstract description 17
- 238000012795 verification Methods 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 6
- 239000002689 soil Substances 0.000 claims description 31
- 230000008595 infiltration Effects 0.000 claims description 27
- 238000001764 infiltration Methods 0.000 claims description 27
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 26
- 238000004891 communication Methods 0.000 claims description 14
- 230000008859 change Effects 0.000 claims description 13
- 239000011148 porous material Substances 0.000 claims description 12
- 230000001133 acceleration Effects 0.000 claims description 10
- 238000000638 solvent extraction Methods 0.000 claims description 10
- 230000035699 permeability Effects 0.000 claims description 9
- 230000000007 visual effect Effects 0.000 claims description 7
- 230000007423 decrease Effects 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 239000003673 groundwater Substances 0.000 claims description 5
- 229920006395 saturated elastomer Polymers 0.000 claims description 4
- 238000009933 burial Methods 0.000 claims description 3
- 230000000295 complement effect Effects 0.000 claims description 3
- 238000009792 diffusion process Methods 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 abstract description 9
- 230000006870 function Effects 0.000 description 13
- 238000010586 diagram Methods 0.000 description 10
- 230000001052 transient effect Effects 0.000 description 6
- 230000006872 improvement Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 231100000817 safety factor Toxicity 0.000 description 3
- 238000007619 statistical method Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A50/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE in human health protection, e.g. against extreme weather
Abstract
The invention belongs to the technical field of rainfall-induced landslide prediction, and discloses a parallel optimization method and a system for TRIGRS of a rainfall-induced landslide model, wherein the TRIGRS2.1 calculation part is optimized based on OpenMP+MPI: on the basis of dividing raster data by MPI, dividing the raster data responsible for each process again by using OpenMP, and realizing thread-level parallelism; using MPI to respectively parallelize a file writing part in the main program and a subprogram rnoff part for calculating runoff penetration and outputting actual penetration rate; and verifying the validity of the parallel strategy through three groups of different data set experiments, and visually displaying the minimum safety coefficient of the verification area. The parallel strategy provided by the invention greatly improves the running speed, has obvious optimizing effect, and has better expandability in processing large-scale data compared with the original model.
Description
Technical Field
The invention belongs to the technical field of rainfall-induced landslide prediction, relates to a parallel optimization method of a rainfall-type landslide model TRIGRS, and particularly relates to a parallel optimization method and system of the rainfall-type landslide model TRIGRS based on OpenMP and MPI.
Background
At present, landslide refers to that a rock-soil body on a side slope loses an original stable state under the action of gravity due to natural or artificial influence, and slides downwards along the side slope and falls off. Landslide is one of the main geological disasters, frequently occurs and has burstiness in the global scope, and is widely distributed in hills, hills and hills. The global landslide area reaches 370 ten thousand square kilometers, and about 5% of the world population is at risk of landslide. Many factors causing landslide, such as heavy rainfall, earthquake, human activity, etc., are involved, and rainfall has become a major factor, and in China, the occurrence frequency of rainfall-induced landslide is 60% of the historical occurrence frequency of landslide. Therefore, it is important to predict, prevent and control rainfall induced landslide.
At present, research methods for predicting rainfall landslide can be divided into three types: based on experience, based on statistical analysis, kinetic model. Experience-based methods are mainly based on analysis of rainfall characteristics such as: rainfall intensity, duration and accumulated rainfall to obtain a threshold of rainfall intensity that induces landslide, this method can only be used in specific areas and has great subjectivity. The model based on statistical analysis comprises Bayes, retrospective analysis, neural networks and the like, so that subjectivity is reduced to a great extent, but the model is only modeled from a data layer, and the mechanism of landslide generation cannot be explained. The model based on dynamics starts from a physical mechanism of landslide occurrence, and takes geographical, geological and hydrologic characteristics into consideration, so that the real rainfall and slope change process is simulated, the model is more fit with reality, and the predicted result is more accurate. Representative of such models are SHALSTAB (Shallow Landslide Stability Model), SINMAP (Stability Index mapping), TRIGRS (Transient Rainfall Infiltration and Grid-based Regional Slope-Stability Model), which takes into account transient rainfall infiltration and the different phases of rainfall intensity on the basis of the first two models. Park D W et al compared similar models, such as SLIP (Shallow Landslides Instability Prediction), SHALSTAB, SINMAP, LISA, to TRIGRS and found that TRIGRS gave better assessment of slope stability. In view of this, the TRIGRS model is widely used by students at home and abroad in various rainfall landslide researches. Kimd et al studied a primary landslide due to heavy rainfall in 1998 in a certain mountain area of korea using TRIGRS, described the course of the safety factor of each divided small area with rainfall intensity using the model, and compared the landslide area predicted by TRIGRS with the area where landslide actually occurs, found that the two reached a matching degree of 64.1%. Saadatkhah N et al take landslide of Hulu Kelang in 2008 and 2009 as an example, calculate the safety factor of the area by TRIGRS, and explore the sensitive area where landslide occurs. Wang J et al combine the statistical moment point estimation method with the TRIGRS model to draw the conclusion that the rainfall has different effects on the stability of gentle and steep slopes. Sugiarti K et al discusses the current uses, deficiencies and improvements of TRIGRS and illustrates a few examples.
The regional slope stability calculation model (TRIGRS) of transient rainfall infiltration is a rainfall landslide model widely applied, but because the calculation of TRIGRS is based on raster data with large data volume and complex numerical simulation calculation is required, the whole calculation process takes a long time and cannot meet the real-time prediction requirement. Under the condition that the current parallel software and hardware technologies such as high-performance computer cluster, parallel programming, distributed computing and the like are continuously developed, the parallel computing of the model is considered, so that the running speed of the model is improved.
In summary, the problems of the prior art are: (1) Existing methods for predicting rainfall landslide based on experience can only be used in specific areas and have great subjectivity.
(2) The existing model based on statistical analysis only models from a data layer, and cannot explain the mechanism of landslide occurrence.
(3) The operation of the existing TRIGRS model is based on raster data with large data volume, and complex numerical simulation operation is required, so that the whole operation process takes a long time, and the real-time prediction requirement cannot be met.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides a parallel optimization method and a system for a rainfall landslide model TRIGRS.
The invention discloses a parallel optimization method of a rainfall landslide model TRIGRS, which comprises the following steps:
firstly, after primary division is carried out on raster data by utilizing MPI, the raster data responsible for each process is divided again by using OpenMP, so that thread level parallelism is realized;
using MPI to respectively parallelize a file writing part in the main program and a subprogram rnoff part for calculating runoff penetration and outputting actual penetration rate;
and thirdly, verifying the validity of three groups of different data set parallel strategies, and visually displaying the minimum safety coefficient of the verification area.
In the first step, input, calculation and output are performed by using a TRIGRS parallel program, and raster data is divided again, specifically including:
(1) The main process reads the input file and stores the read data into the corresponding variable and array; then, the main process broadcasts the data to other processes through an MPI_Bcast function, so that all processes have the data required by calculation;
(2) Partitioning the raster unit to enable each process to be responsible for calculating part of raster data, enabling the calculation of each process to be performed in parallel, reading raster data stored in the one-dimensional array in the step (1), partitioning the raster data into idsp and isct array data, enabling the idsp to store the initial position of data in the array, which is responsible for each process, and enabling the isct to store the data size which is responsible for each process; after the calculation of each process is finished, integrating the result into the main process through the MPI_GatherV and the MPI_reduce;
(3) The host process is responsible for writing the results to the output file.
Further, the input data of TRIGRS includes digital elevation, gradient, rainfall intensity grid data, and soil, hydrographic physical parameters and control parameters required by the model;
the TRIGRS comprises an infiltration model, a hydrological model and a slope stability model;
the hydrologic model simulates surface runoff, rainfall which cannot infiltrate in time in the current time directly flows into a downstream grid of the current grid in a surface runoff mode, so that mass conservation of the runoff in each operation time domain is realized;
the infiltration model comprises stable infiltration and instantaneous infiltration, and is used for analyzing landslide occurring when the bottom layer is in saturation or near saturation or landslide occurring when the ground water level is raised due to the infiltration of rainwater in the unsaturated stratum;
and the slope stability model is used for calculating the safety coefficient of each grid in different rainfall stages and predicting the stability of the slope.
Further, the infiltration model calculates two bedrock boundaries: (1) The permeability of the bedrock is the same as that of the soil body (2) the permeability of the bedrock is smaller than that of the soil body, the bedrock boundary of the (2) class is a pore water pressure relation function which changes with different rainfall intensities and time durations at a limited depth, and the expression is as follows:
wherein t is the total time of calculating psi; z is the soil layer thickness in the vertical direction, Z=z/cos alpha, and Z is the soil layer thickness perpendicular to the slope direction; d is the vertical underground water level burial depth measured in a stable state; k (K) s Is the saturation vertical permeability coefficient; i nz The ground infiltration amount corresponding to the rainfall intensity of the nth period; beta =cos 2 α-(I zlt /K s )，I zlt For stable (initial) surface penetration, it is generally available from average rainfall over nearly several weeks or months; d (D) 1 ＝D 0 /cos 2 α，D 0 Is the saturation hydraulic diffusion coefficient (D 0 ＝K s /S s ，S s Water storage coefficient ratio); n is the total number of rainfall duration intervals; h (t-t) n ) Is a sea-Vield step function, t n The duration of rainfall in the nth stage in the rainfall period;
the expression of the ierfc (η) function is:
where erfc (η) is the complementary error function. The first type of calculation process can be referred to the study of Baum R.L.
The slope stability model is combined with the change of groundwater pore water pressure to obtain the expression of the stability of the grid unit body at the z position with different depths, wherein the expression is as follows:
wherein c' is the effective cohesive force of the soil,is the effective internal friction angle of soil, gamma w Is the volume weight of water, gamma s Is the natural volume weight of the soil. ψ (Z, t) is the pore water pressure; according to F s The evaluation result is divided into a high-probability area, a medium-probability area, a low-probability area and a more stable area by 4 grades according to the value; when F s When the value is smaller than 1, the grid is considered to be in a very unstable state and belongs to a high-probability area; when F s When the value is more than 1 and less than 2, the method belongs to a medium-frequency easy-occurrence area; when F s When the value is more than 1.2 and less than 1.5, the method belongs to a low-incidence area; when F s When the value is greater than 1.5, the region is more stable.
In the second step, the file writing part and the subprogram rnoff part in the main program are provided with one do cycle, the cycle number of the do cycle is determined by a parameter nout, the cycle number of the do cycle is determined by nper, and nout is the number of rainfall moments to be output, which are set in the initialized file; nper is the number of rainfall phases;
the MPI is utilized to carry out parallel on the loops, and when the process number of the optimized program is larger than nout, the file writing is the time spent by one loop; when the number of processes is smaller than nout, the file writing time is shortened by several times.
Further, in the third step, the experimental evaluation index includes:
(1) Speed-up ratio
The speed-up ratio is a basic index for measuring the parallelization performance and effect of a program, and the formula is defined as follows:
wherein T is 1 For the running time of the parallel preprogrammes, T p Designating the running time of parallel program when there are P processors;
(2) Efficiency of
The efficiency is a parallel performance metric derived from the acceleration ratio, defined as:
the efficiency is reflected by the degree of processor utilization of the participation calculation when the parallel program solves the problem, compared with the cost of communication in synchronization; as the number of processes increases, the time taken for communication increases, and the efficiency of the parallel process decreases.
Another object of the present invention is to provide a parallel optimization system of the rainfall landslide model TRIGRS, which includes:
the thread level parallel dividing module is used for dividing raster data for the first time by utilizing MPI, dividing the raster data responsible for each process again by using OpenMP, and realizing the parallel of thread levels;
the parallel processing module is used for respectively parallelizing a file writing part in the main program and a subprogram rnoff part for calculating runoff penetration and outputting actual penetration rate by using MPI;
and the visual display module is used for verifying the validity of the parallel strategies of the three different data sets and visually displaying the minimum safety coefficient of the verification area.
Further, the thread level parallel partitioning module includes:
for an input module, the main process reads an input file and stores read data into a corresponding variable and array; the main process broadcasts data to other processes through an MPI_Bcast function, and all processes have data required by calculation;
the calculation module is used for partitioning the grid units, so that each process is responsible for calculating part of grid data, the calculation of each process is performed in parallel, the grid data stored in the one-dimensional array by the read input module is partitioned, and after the calculation of each process is finished, the result is integrated into the main process through MPI_GatherV and MPI_reduce;
and the output module is responsible for writing the result into the output file by the main process.
It is another object of the present invention to provide a computer program product stored on a computer readable medium, comprising a computer readable program for providing a user input interface for implementing a parallel optimization method of a rainfall landslide model TRIGRS when executed on an electronic device.
Another object of the present invention is to provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the parallel optimization method of rainfall landslide model TRIGRS.
In summary, the invention has the advantages and positive effects that: TRIGRS2.1 is changed from a serial program to an MPI parallel program, where the real use of the MPI is parallel to the computation part, the file reading and writing part is essentially performed in series by a process. Meanwhile, in the MPI process communication, there are many redundant data sending and receiving, which affect the communication time and occupy many unnecessary memory spaces. Based on the parallel implementation of the current version and the existing defects, the invention provides a parallel strategy.
According to the parallel optimization method for the rainfall landslide model TRIGRS, the latest parallel version of the TRIGRS is further optimized by utilizing the MPI and OpenMP hybrid programming technology. The invention mainly concentrates the calculation and file writing part of the optimization model, and experiments are respectively carried out on three groups of different data sets so as to evaluate the optimization effect. Experimental results show that the running speed is greatly improved, the optimization effect is obvious, and compared with the original model, the optimized model has better expandability in processing large-scale data. Meanwhile, the input and output results of the model are visualized by using ArcGIS software.
Through experiments of three groups of different data sets, the parallel strategy provided by the invention not only obtains good acceleration effect, but also reduces the communication traffic of the program due to code optimization, so that the original edition which cannot run a large data set due to memory limitation can be normally operated. Besides verifying the optimization effect, the method carries out visual display on the minimum safety coefficient of the region on the Shenzhen data set, and more intuitively sees the application scene of TRIGRS in reality.
Drawings
Fig. 1 is a flowchart of a parallel optimization method of a rainfall landslide model TRIGRS provided by an embodiment of the invention.
Fig. 2 is a parallel optimization system diagram of a rainfall landslide model TRIGRS provided by an embodiment of the invention.
In the figure: 1. a thread level parallel dividing module; 1-1, an input module; 1-2, a calculation module; 1-3, an output module; 2. a parallel processing module; 3. and a visual display module.
Fig. 3 is a schematic diagram of a TRIGRS framework according to an embodiment of the present invention.
Fig. 4 is a schematic block diagram of a TRIGRS runtime according to an embodiment of the present invention.
Fig. 5 is a schematic diagram of an openmp+mpi hybrid programming model according to an embodiment of the present invention.
FIG. 6 is a schematic view of a topographical image, DEM, and slope of Shenzhen city according to an embodiment of the present invention.
Fig. 7 is a schematic diagram showing experimental performance of the data set 1 provided in the embodiment of the present invention at different process numbers.
FIG. 8 is a diagram of the relationship between the running time of the data set 1 and the CPU core number according to the embodiment of the present invention.
Fig. 9 is a schematic diagram of the relationship between the running time and the thread number of the data set 2 (p=10 and 16) according to the embodiment of the present invention.
FIG. 10 is a schematic diagram of the total program running time, acceleration ratio and efficiency before and after optimization according to an embodiment of the present invention.
FIG. 11 is a diagram showing the relationship between the total running time of the data set 2 and the CPU core number according to the embodiment of the present invention.
FIG. 12 is a graph showing minimum safety factors at 0hr, 4hr, and 10hr of rainfall provided by the embodiment of the present invention.
FIG. 13 is a diagram of the relationship between the running time of the data set 3 and the number of processes according to the embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Aiming at the problems existing in the prior art, the invention provides a parallel optimization method of a rainfall landslide model TRIGRS, and the invention is described in detail below with reference to the accompanying drawings.
As shown in fig. 1, the parallel optimization method for the rainfall landslide model TRIGRS provided by the embodiment of the invention comprises the following steps:
s101, calculating partial optimization by TRIGRS2.1 based on OpenMP: on the basis that the MPI is used for dividing the raster data, openMP is used for dividing the raster data responsible for each process again, and thread-level parallelism is achieved.
S102, using MPI, the file writing part in the main program and the subprogram rnoff part for calculating runoff penetration and outputting the actual penetration rate are respectively parallel.
And S103, verifying the validity of the parallel strategy through three groups of different data set experiments, and visually displaying the minimum safety coefficient of the verification area.
As shown in fig. 2, the parallel optimization system of the rainfall landslide model TRIGRS provided by the present invention includes:
and the thread level parallel dividing module 1 is used for dividing raster data for the first time by utilizing MPI, and dividing the raster data responsible for each process again by using OpenMP so as to realize the parallel of the thread level.
The parallel processing module 2 is used for respectively parallelizing a file writing part in the main program and a subprogram rnoff part for calculating runoff penetration and outputting actual penetration rate by using MPI.
And the visual display module 3 is used for verifying the validity of the parallel strategies of the three different data sets and visually displaying the minimum safety coefficient of the verification area.
The thread level parallel division module 1 includes:
the input module 1-1, the main process reads the input file and stores the read data into the corresponding variable and array; the main process broadcasts the data to other processes through the MPI_Bcast function, and all processes have the data required by calculation.
And the calculation module 1-2 is used for partitioning the grid units, so that each process is responsible for calculating part of grid data, the calculation of each process is performed in parallel, the grid data stored in the one-dimensional array by the reading input module is partitioned, and after the calculation of each process is finished, the result is integrated into the main process through MPI_GatherV and MPI_reduce.
The output module 1-3 is responsible for writing the result to the output file by the host process.
The invention is further described below with reference to examples.
1、TRIGRS
Transient Rainfall Infiltration and Grid-based Regional Slope-Stability Model (TRIGRS, regional slope Stability calculation Model for transient rainfall infiltration) is a grid unit-based rainfall induced slope Stability calculation Model written in Fortran language by the united states geological survey. The model is continuously improved, and the latest current version is TRIGRS 2.1. The biggest improvement of TRIGRS2.1 is to parallelize the use of MPI in addition to correcting code errors, formulas, etc. in the previous version, thus shortening the runtime of the model.
TRIGRS can simulate and analyze the change of transient pore water pressure and the change of slope safety coefficient caused by the real rainfall process of the change of rainfall intensity of soil in a saturated state and an unsaturated state, and the whole method can be divided into: the input module, the calculation module and the output module are arranged in the framework shown in fig. 3. The input data of TRIGRS are divided into two categories: one is grid data such as DEM (Digital Elevation Model, digital elevation), gradient, rainfall intensity and the like, and the other is physical parameters such as soil property, hydrology and the like and control parameters required by a model. TRIGRS reads both input data via an initialization file. After the TRIGRS reads the input data, the TRIGRS enters a calculation part, and finally, the calculated results such as pore water pressure, safety coefficient and the like are written into an output file.
The TRIGRS consists of three sub-models, namely an infiltration model, a hydrological model and a slope stability model. The hydrologic model simulates surface runoff, and TRIGRS assumes that rainfall which cannot infiltrate in time in the current time directly flows into a downstream grid of the current grid in a surface runoff mode, so that mass conservation of the runoff in each operation time domain is ensured.
The infiltration model comprises stable infiltration and instantaneous infiltration, and the sub-model is a change model of Baum, which is based on the instantaneous rainfall infiltration model proposed by Iverson, and is coupled with rainfall infiltration and instantaneous pore water pressure, so that the model can analyze landslide occurring when a bottom layer is saturated or nearly saturated, and landslide occurring when an underground water level is raised due to the rainwater infiltration of an unsaturated stratum. TRIGRS can calculate two bedrock boundaries: (1) The permeability of the bedrock is the same as that of the soil body (2) the permeability of the bedrock is smaller than that of the soil body, wherein the second type bedrock boundary is a pore water pressure relation function which changes with different rainfall intensities and time periods at a limited depth, and the expression is as follows:
wherein t is the total time of calculating psi; z is the soil layer thickness in the vertical direction, Z=z/cos alpha, and Z is the soil layer thickness perpendicular to the slope direction; d is the vertical underground water level burial depth measured in a stable state; k (K) s Is the saturation vertical permeability coefficient; i nz The ground infiltration amount corresponding to the rainfall intensity of the nth period; beta = cos 2 α-(I zlt /K s )，I zlt For stable (initial) surface penetration, it is generally available from average rainfall over nearly several weeks or months; d (D) 1 ＝D 0 /cos 2 α，D 0 Is the saturation hydraulic diffusion coefficient (D 0 ＝K s /S s ，S s Water storage coefficient ratio); n is the total number of rainfall duration intervals; h (t-t) n ) Is a sea-Vield step function, t n Is the rainfall duration of the nth stage in the rainfall period.
The expression of the ierfc (η) function is:
where erfc (η) is the complementary error function. The first type of calculation process can be referred to the study of Baum R.L.
The slope stability model assumes an infinite slope, and in the whole rainfall process, the model calculates the safety coefficient of each grid in different rainfall stages, so as to predict the stability of the slope. Based on Moire-Coulomb destruction criteria and combined with the change of groundwater pore water pressure, the expression of grid unit stability at different depths z is obtained as follows:
wherein c' is the effective cohesive force of the soil,is the effective internal friction angle of soil, gamma w Is the volume weight of water, gamma s Is the natural volume weight of the soil. ψ (Z, t) is the pore water pressure. Binding to related studies according to F s The value size divides the evaluation result into 4 grades of high-incidence area, medium-incidence area, low-incidence area and more stable area. When F s When the value is smaller than 1, the grid is considered to be in a very unstable state and belongs to a high-probability area; when F s When the value is more than 1 and less than 2, the method belongs to a medium-frequency easy-occurrence area; when F s When the value is more than 1.2 and less than 1.5, the method belongs to a low-incidence area; when F s When the value is greater than 1.5, the region is more stable.
1.2TRIGRS 2.1 code Structure
Alvioli uses MPI to parallelize TRIGRS to form TRIGRS version 2.1. The parallel program corresponds to the input, calculation and output modules, and the corresponding running time is also divided into three parts, as shown in fig. 4.
For the input module, firstly, the main process reads an input file and stores the read data into a corresponding variable and array; the master process then broadcasts the data to the other processes through the mpi_bcast function so that all processes have the data needed for the computation.
For the calculation module, since the calculation of TRIGRS is based on the grid units, the calculation of each grid is independent and does not affect each other, so that the grid units are partitioned, each process is responsible for calculating part of grid data, and the calculation of each process is performed in parallel. In the input module, the read raster data has been stored in a one-dimensional array, so the partitioning of the raster data is effectively a division of the array. TRIGRS2.1 implements data blocking by two arrays idsp and isct, the former storing the starting position of the data in the array that each process is responsible for, and the latter storing the size of the data volume that each process is responsible for. Finally, after each process calculation is completed, the results are integrated into the master process by MPI_GatherV and MPI_reduce.
For the output module, the main process is responsible for writing the results to the output file.
In summary, TRIGRS2.1 changes from a serial program to an MPI parallel program, where the real use of MPI in parallel is the computation part, and the file reading and writing parts are essentially performed in series by one process. Meanwhile, in the MPI process communication, there are many redundant data sending and receiving, which affect the communication time and occupy many unnecessary memory spaces. Based on the parallel implementation of the current version and the defects existing, the invention provides own parallel strategy.
2. Parallel policy
2.1OpenMP parallelism
The invention uses OpenMP to optimize the calculation part of TRIGRS2.1, namely, on the basis of dividing raster data by MPI, the invention uses OpenMP to divide the raster data responsible for each process again, thus realizing thread-level parallelism, as shown in figure 5. In OpenMP parallelism, the memory of a process is shared with threads belonging to the process, so that the extra consumption of communication and space overhead is also small.
2.2MPI parallelism
The invention uses MPI to parallel two parts of the program, namely a file writing part in the main program and a subprogram rnoff for calculating runoff penetration and outputting actual penetration rate. Both have one do cycle, the number of which is determined by the parameter nout, the number of which is determined by nper, and nout is the number of rainfall moments set in the initialization file to be output; nper is the number of rainfall phases. The invention uses MPI to parallelize the loops, and the pseudo code after the file writing part of the main program is parallelized is as follows.
When the number of processes of the optimized program is larger than nout, the file writing is the time spent by one cycle; when the number of processes is smaller than nout, the file writing time is also shortened by several times. The manner and effect achieved in subroutine rnoff is similar. In parallel with MPI, the present invention finds that there are many unnecessary MPI-Reduce operations in MPI communication, the same objective is achieved after they are replaced with MPI_GatherV, and data traffic is reduced.
3. Experiment and results
3.1 Experimental data
The invention shares three groups of data for experiments. The DEM data for data set 1 and data set 3 are from SRTM 1 The downloaded data set 2 is data of Shenzhen city in a certain rectangular area in the middle of China, the Shenzhen global terrain is high in southeast and low in northwest, the soil is mostly low mountain, gentle bench and terraced hills, the soil texture is mainly sandy loam and light loam, rainfall is rich, and landslide occurrence possibility is high. DEM data of Shenzhen city is downloaded from water through map downloading software, and specific information of three data sets is shown in table 1.
Table 1 dataset information
The physical parameters such as water, soil, rainfall and the like used by the data set 2 take the real data of Shenzhen city as references, so that the result of program operation has certain practical significance. Meanwhile, compared with the data sets 1 and 3 and 2, all valid values are in the raster data, and the data set 2 is an irregular area like Shenzhen, so that a plurality of invalid values are contained in the raster data, and the complexity of operation is increased. The present invention will now be described in detail with respect to various data related to the data set 2. The soil and water parameters required for TRIGRS input are shown in table 2.
Table 2 water and soil parameters for data set 2
The rainfall data are used as references for 7 months 21 and 22 in Shenzhen Longpost district, and are divided into 5 rainfall phases with different rainfall densities, and each phase is 2 hours long, as shown in table 3.
Table 3 5 rainfall density for the rainfall phases
Based on DEM data, the ArcGIS can be used for generating gradient and flow direction files required by TRIGRS input, and FIGS. 6 (A), (B) and (C) respectively show the topography image, DEM and gradient conditions of Shenzhen city, and the rest of input files and model control parameters are generated according to the setting requirements.
3.2 Experimental evaluation index
3.2.1 speed ratio
The speed-up ratio is a basic index for measuring the parallelization performance and effect of a program, and the formula is defined as follows:
wherein T is 1 For the running time of the parallel preprogrammes, T p The running time of parallel program when P processors exist is indicated, and obviously, the larger the speed-up ratio is, the better the parallel effect is.
3.2.2 efficiency
The efficiency is a parallel performance metric derived from the acceleration ratio, defined as:
the efficiency is represented by the degree to which the processors involved in the computation are utilized in solving the problem with the parallel program, as opposed to the expense of communicating in synchronization. Generally, as the number of processes increases, the time taken for communication increases, and the efficiency of parallel processes decreases gradually.
3.3 experimental results
The invention completes all experiments in BigData partition of Tianhe No. 2, the Guangzhou super computing platform, and users can use 64 nodes at most under the platform, each node is provided with 2 Intel (R) Xeon (R) CPUE5-2692 v2@2.20GHz, and each CPU has 12 cores. Each node installs Red Hat Enterprise Linux Server release 6.5.5 (Santiago) operating system, some other necessary software is: gcc-4.8.5,Mpich 3.2.1,gsl/2.4-gcc-4.8.5.
3.3.1 data set 1 experimental results
On the data set 1, the invention respectively explores the change situations of calculation time, file writing time, file reading time and total time before and after optimization, as shown in fig. 7 (a). Overall, each part of time after optimization is significantly shortened compared to before optimization. After the number of processes reaches 12, the running total time and the calculation time of the program before optimization are reduced by a small extent, because the faster the part participating in calculation is executed with the increase of the number of processes, the larger the time occupation ratio of communication is at the same time, so that the overall optimization effect is poor. The optimized program realizes the parallel on the threads, and the calculation part is greatly reduced, so that the calculation time and the total time of the optimized program are very short from the beginning, and the gradual descending trend is kept all the time. The file writing time before optimization is irrelevant to the number of processes, and the file writing time after optimization is reduced by times due to parallel writing of multiple processes. Fig. 8 shows the change of the time of each part before and after the optimization along with the number of cores of the CPU, and the overall change trend is the same as that of the process, and it is noted that the program before the optimization can only detect 64 cores due to the limitation of the single-node memory.
The present invention shows the optimization effect with fig. 7 (B). It can be seen that the acceleration ratio shows a tendency to rise and then fall, the maximum acceleration ratio is approximately 9, is reached at a number of steps of 8, and then slowly falls, but remains around 5.
As shown in fig. 7 (C), the efficiency of the optimization is continuously decreasing, eventually approaching zero, which also coincides with the change in run time and acceleration ratio, and when a certain number of processes is exceeded, the number of processes is increased, and the added additional optimization effect is small.
3.3.2 data set 2 experimental results
In order to determine the optimal thread number setting in data set 2, the present invention tests the total time of program operation for different thread numbers when the thread numbers are 10 and 16, respectively, and fig. 9 is an experimental result. As can be seen from the figure, the running time is at a minimum when the thread count is set to 8, and therefore, in an experiment for testing the relation between the running time and the process count, the thread count is set to 8. The same method is also used in data set 1 to find the optimal thread count for data set 1 to be set to 16.
Fig. 10 shows the total program running time, acceleration ratio and efficiency before and after the optimization. As can be seen from fig. 10 (a), the process number 6 is a demarcation point, and when the process number is smaller than 6, the total running time before and after the optimization is rapidly reduced with the increase of the process number; when the number of processes is greater than 6, the drop is very slow. As can be seen from fig. 10 (B), the speed is increased by at most approximately 6 times after the optimization compared with the speed before the optimization, and the speed-up ratio exhibits a state stabilized around 5. From fig. 10 (C), the efficiency of the optimization decreases as the number of processes increases, also because as the number of processes increases, the number of computation portions each processor takes charge of decreases, while the amount of inter-process traffic increases, which takes much time. Compared with the data set 1, the total time occupied by the calculation time in the data set 2 is smaller, the performance improvement brought by the optimization calculation part is smaller, and the final speed-up ratio is smaller.
FIG. 11 shows the variation of the running time with the number of CPU cores before and after the optimization, the running time variation trend of the running time variation is basically the same, the running time variation gradually decreases with the increase of the number of CPU cores, and the reduction is also gradually reduced.
According to the invention, the safety coefficient output by the TRIGRS model is visualized by using ArcGIS software, as shown in fig. 12, the safety coefficient of each grid unit corresponds to the safety coefficient before rainfall, 4 hours before rainfall and 10 hours after rainfall from left to right, so that the change condition of the stability of the area can be seen. From the visual result, the area safety coefficient is not obviously changed in the area, because (1) the area is mainly a high-incidence area and a relatively stable area of the slope before rainfall, and the medium-incidence area and the low-incidence area are few; (2) the safety level of the high-probability area is not changed after rainfall; (3) According to the initial parameter calculation, the safety coefficient of the area is larger than that of the stable area, and after rainfall, the safety coefficient obtained by TRIGRS calculation is reduced to a certain extent, but is still larger than 1.5, the safety level is not changed, and the area still belongs to the stable area.
3.3.3 dataset 3 experimental results
The data set 3 is used to verify the scalability of the optimized program. Because of MPI_reduce operation of a plurality of arrays with the multiple of the total number of grids exists in the program before optimization, the program cannot normally run in operation due to the limitation of a memory. The optimized program can normally run the data set of the data volume level due to the MPI improvement. As shown in FIG. 13, the variation trend of the two times is quite consistent, because the time spent writing the file is basically unchanged when the number of processes exceeds the number of output files, namely, the file is written for one process.
4. The invention utilizes the OpenMP and MPI mixed programming technology to further optimize the TRIGRS model. Through experiments of three groups of different data sets, the parallel strategy provided by the invention not only obtains good acceleration effect, but also reduces the communication traffic of the program due to code optimization, so that the original edition which cannot run a large data set due to memory limitation can be normally operated. Besides verifying the optimization effect, the method carries out visual display on the minimum safety coefficient of the region on the Shenzhen data set, and more intuitively sees the application scene of TRIGRS in reality.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When used in whole or in part, is implemented in the form of a computer program product comprising one or more computer instructions. When loaded or executed on a computer, produces a flow or function in accordance with embodiments of the present invention, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital Subscriber Line (DSL), or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid State Disk (SSD)), etc.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (8)
1. The parallel optimization method of the rainfall type landslide model TRIGRS is characterized by comprising the following steps of:
firstly, dividing raster data for the first time by utilizing MPI, dividing the raster data responsible for each process again by using OpenMP, and carrying out parallel thread level;
using MPI to respectively parallelize a file writing part in the main program and a subprogram rnoff part for calculating runoff penetration and outputting actual penetration rate;
step three, verifying the validity of three groups of different data set parallel strategies, and visually displaying the minimum safety coefficient of the verification area;
in the second step, the file writing part in the main program and the subprogram rnoff part are provided with a do cycle, the cycle number of the do cycle is determined by a parameter nout, the cycle number of the do cycle is determined by nper, and nout is the number of rainfall moments which are set in the initialized file and need to be output; nper is the number of rainfall phases;
the MPI is utilized to carry out parallel on the loops, and when the process number of the optimized program is larger than nout, the file writing is the time spent by one loop; when the number of processes is smaller than nout, the file writing time is shortened by several times of the number of processes.
2. The parallel optimization method of rainfall landslide model TRIGRS according to claim 1, wherein in the first step, input, calculation and output are performed by using a TRIGRS parallel program, and raster data is subdivided, specifically including:
(1) The main process reads the input file and stores the read data into the corresponding variable and array; then, the main process broadcasts the data to other processes through an MPI_Bcast function, so that all processes have the data required by calculation;
(2) Partitioning the raster unit to enable each process to be responsible for calculating part of raster data, enabling the calculation of each process to be performed in parallel, reading raster data stored in the one-dimensional array in the step (1), partitioning the raster data into idsp and isct array data, enabling the idsp to store the initial position of data in the array, which is responsible for each process, and enabling the isct to store the data size which is responsible for each process; after the calculation of each process is finished, integrating the result into the main process through the MPI_GatherV and the MPI_reduce;
(3) The host process is responsible for writing the results to the output file.
3. A parallel optimization method of a rainfall landslide model TRIGRS as claimed in claim 2, wherein the input data of TRIGRS includes digital elevation, gradient, rainfall intensity grid data, and soil, hydrologic physical parameters and control parameters required by the model;
the TRIGRS comprises an infiltration model, a hydrological model and a slope stability model;
the hydrologic model simulates surface runoff, rainfall which cannot infiltrate in time in the current time directly flows into a downstream grid of the current grid in a surface runoff mode, so that mass conservation of the runoff in each operation time domain is realized;
the infiltration model comprises stable infiltration and instantaneous infiltration, and is used for analyzing landslide occurring when the bottom layer is in saturation or near saturation or landslide occurring when the ground water level is raised due to the infiltration of rainwater in the unsaturated stratum;
and the slope stability model is used for calculating the safety coefficient of each grid in different rainfall stages and predicting the stability of the slope.
4. A parallel optimization method for a rainfall landslide model TRIGRS as claimed in claim 3, wherein the infiltration model calculates two bedrock boundaries: (1) The permeability of the bedrock is the same as that of the soil body (2) the permeability of the bedrock is smaller than that of the soil body, the boundary of the bedrock of the (2) type is a pore water pressure relation function which changes with different rainfall intensities and time durations at a limited depth, and the expression is as follows:
wherein t is the total time of calculating psi; z is the soil layer thickness in the vertical direction, Z=z/cos alpha, and Z is the soil layer thickness perpendicular to the slope direction; d is the vertical underground water level burial depth measured in a stable state; k (K) s Is the saturation vertical permeability coefficient; i nz The ground infiltration amount corresponding to the rainfall intensity of the nth period; beta = cos 2 α-(I zlt /K s )，I zlt To stabilize the initial surface penetration, it is obtained from an average rainfall over nearly several weeks or months; d (D) 1 ＝D 0 /cos 2 α，D 0 Is the saturated hydraulic diffusion coefficient, D 0 ＝K s /S s ，S s The water storage coefficient is the ratio; n is the total number of rainfall duration intervals; h (t-t) n ) Is a sea-Vield step function, t n The duration of rainfall in the nth stage in the rainfall period;
the expression of the ierfc (η) function is:
wherein erfc (η) is a complementary error function, the calculation of the (1) st species may be referred to the Baum R.L study,
the slope stability model is combined with the change of groundwater pore water pressure to obtain the expression of the stability of the grid unit body at the z position with different depths, wherein the expression is as follows:
wherein c' is the effective cohesive force of the soil,is the effective internal friction angle of soil, gamma w Is the volume weight of water, gamma s Is the natural volume weight of the soil, and psi (Z, t) is the pore water pressure; according to F s The evaluation result is divided into a high-probability area, a medium-probability area, a low-probability area and a more stable area by 4 grades according to the value; when F s When the value is smaller than 1, the grid is considered to be in a very unstable state and belongs to a high-probability area; when F s When the value is more than 1 and less than 2, the method belongs to a medium-frequency easy-occurrence area; when F s When the value is more than 1.2 and less than 1.5, the method belongs to a low-incidence area; when F s When the value is greater than 1.5, the region is more stable.
5. The parallel optimization method of rainfall landslide model TRIGRS of claim 1, wherein in the third step, the experimental evaluation index comprises:
(1) Speed-up ratio
The speed-up ratio is a basic index for measuring the parallelization performance and effect of a program, and the formula is defined as follows:
wherein T is 1 For the running time of the parallel preprogrammes, T p Designating the running time of parallel program when there are P processors;
(2) Efficiency of
The efficiency is a parallel performance metric derived from the acceleration ratio, defined as:
the efficiency is reflected by the degree of processor utilization of the participation calculation when the parallel program solves the problem, compared with the cost of communication in synchronization; as the number of processes increases, the time taken for communication increases, and the efficiency of the parallel process decreases.
6. A parallel optimization system of a rainfall type landslide model TRIGRS for implementing the parallel optimization method of the rainfall type landslide model TRIGRS according to any one of claims 1 to 5, characterized in that the parallel optimization system of the rainfall type landslide model TRIGRS comprises:
the thread level parallel dividing module is used for dividing raster data for the first time by utilizing MPI, dividing the raster data responsible for each process again by using OpenMP, and realizing the parallel of thread levels;
the parallel processing module is used for respectively parallelizing a file writing part in the main program and a subprogram rnoff part for calculating runoff penetration and outputting actual penetration rate by using MPI; the file writing part and the subprogram rnoff part in the main program are provided with one do cycle, the cycle number of the do cycle is determined by a parameter nout, the cycle number of the do cycle is determined by nper, and nout is the number of rainfall moments which are set in an initialization file and need to be output; nper is the number of rainfall phases;
the MPI is utilized to carry out parallel on the loops, and when the process number of the optimized program is larger than nout, the file writing is the time spent by one loop; when the process number is smaller than nout, the file writing time is shortened by several times of the process number;
and the visual display module is used for verifying the validity of the parallel strategies of the three different data sets and visually displaying the minimum safety coefficient of the verification area.
7. The parallel optimization system of rainfall landslide model TRIGRS of claim 6 wherein the thread-level parallel partitioning module comprises:
for an input module, the main process reads an input file and stores read data into a corresponding variable and array; the main process broadcasts data to other processes through an MPI_Bcast function, and all processes have data required by calculation;
the calculation module is used for partitioning the grid units, so that each process is responsible for calculating part of grid data, the calculation of each process is performed in parallel, the grid data stored in the one-dimensional array by the read input module is partitioned, and after the calculation of each process is finished, the result is integrated into the main process through MPI_GatherV and MPI_reduce;
and the output module is responsible for writing the result into the output file by the main process.
8. A computer readable storage medium storing instructions that, when run on a computer, cause the computer to perform the parallel optimization method of rainfall landslide model TRIGRS according to any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010442518.6A CN111625954B (en) | 2020-05-22 | 2020-05-22 | Parallel optimization method and system for rainfall landslide model TRIGRS |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010442518.6A CN111625954B (en) | 2020-05-22 | 2020-05-22 | Parallel optimization method and system for rainfall landslide model TRIGRS |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111625954A CN111625954A (en) | 2020-09-04 |
| CN111625954B true CN111625954B (en) | 2023-10-27 |
Family
ID=72272126
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010442518.6A Active CN111625954B (en) | 2020-05-22 | 2020-05-22 | Parallel optimization method and system for rainfall landslide model TRIGRS |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111625954B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113435140B (en) * | 2021-07-22 | 2022-03-29 | 河海大学 | Rainfall infiltration boundary processing method suitable for complex rainfall time sequence |
| CN114236095B (en) * | 2021-12-02 | 2024-03-19 | 山东高速集团四川乐宜公路有限公司 | Regional grading early warning method for rainfall induced landslide along mountain expressway |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008144960A1 (en) * | 2007-05-31 | 2008-12-04 | Intel Coporation | Method and apparatus for mpi program optimization |
| KR20130039191A (en) * | 2011-10-11 | 2013-04-19 | 한국수자원공사 | Grid based long term rainfall runoff model for large scale watersheds |
| CN103399344A (en) * | 2013-07-24 | 2013-11-20 | 北京师范大学 | Prediction method for predicating collapse disaster position after earthquake |
| WO2018016703A1 (en) * | 2016-07-18 | 2018-01-25 | 연세대학교 산학협력단 | Wireless sensor network measurement system and measurement method for monitoring, forecasting and alarming landslide |
| CN108776851A (en) * | 2018-06-11 | 2018-11-09 | 广州地理研究所 | A kind of shallow failure disaster alarm Threshold that heavy rain induces |
-
2020
- 2020-05-22 CN CN202010442518.6A patent/CN111625954B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008144960A1 (en) * | 2007-05-31 | 2008-12-04 | Intel Coporation | Method and apparatus for mpi program optimization |
| KR20130039191A (en) * | 2011-10-11 | 2013-04-19 | 한국수자원공사 | Grid based long term rainfall runoff model for large scale watersheds |
| CN103399344A (en) * | 2013-07-24 | 2013-11-20 | 北京师范大学 | Prediction method for predicating collapse disaster position after earthquake |
| WO2018016703A1 (en) * | 2016-07-18 | 2018-01-25 | 연세대학교 산학협력단 | Wireless sensor network measurement system and measurement method for monitoring, forecasting and alarming landslide |
| CN108776851A (en) * | 2018-06-11 | 2018-11-09 | 广州地理研究所 | A kind of shallow failure disaster alarm Threshold that heavy rain induces |
Non-Patent Citations (10)
| Title |
|---|
| 《Parallelization of the TRIGRS model for rainfall-induced landslides using the message passing interface》;M. Alvioli.etc;《Environmental Modelling & Software》;第122-135页 * |
| M. Alvioli.etc.《Parallelization of the TRIGRS model for rainfall-induced landslides using the message passing interface》.《Environmental Modelling & Software》.2016,第122-135页. * |
| TRIGRS模型预测降雨型浅层滑坡的应用性评价;徐沅鑫等;第35届中国气象学会年会;全文 * |
| 刘磊;殷坤龙;王佳佳;张俊;黄发明.降雨影响下的区域滑坡危险性动态评价研究――以三峡库区万州主城区为例.岩石力学与工程学报.2016,第35卷(第03期),第558-569页. * |
| 吴李泉;张锋;凌贤长;朱占元;王立娜;高霞.强降雨条件下浙江武义平头村山体高边坡稳定性分析.岩石力学与工程学报.2009,(第06期),全文. * |
| 徐沅鑫;郭海燕;马振峰.TRIGRS模型预测降雨型浅层滑坡的应用性评价.高原气象.2018,(第03期),全文. * |
| 文海家;张岩岩;付红梅;谢朋;胡晶.降雨型滑坡失稳机理及稳定性评价方法研究进展.中国公路学报.2018,(第02期),全文. * |
| 邹祖银;朱占元;张锋;陈婷婷;张自兴;.连续降雨条件下某震后高边坡稳定性分析.地震工程学报.2016,(第04期),全文. * |
| 降雨入渗作用下秭归向斜核部南段斜坡稳定性评价;齐信;刘广宁;黄波林;王世昌;黄长生;;水土保持通报(第03期);全文 * |
| 降雨型滑坡的集合预报模型及其初步应用的试验研究;陈悦丽;陈德辉;李泽椿;吴亚丽;黄俊宝;;大气科学(03);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111625954A (en) | 2020-09-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xia et al. | A full-scale fluvial flood modelling framework based on a high-performance integrated hydrodynamic modelling system (HiPIMS) | |
| Hamman et al. | The Variable Infiltration Capacity model version 5 (VIC-5): Infrastructure improvements for new applications and reproducibility | |
| Fleury et al. | Modelling of the functioning of karst aquifers with a reservoir model: Application to Fontaine de Vaucluse (South of France) | |
| Kalyanapu et al. | Assessment of GPU computational enhancement to a 2D flood model | |
| Leao et al. | Simulating urban growth in a developing nation’s region using a cellular automata-based model | |
| WO2022007398A1 (en) | Unstructured grid flood simulation system based on gpu acceleration thechnology | |
| CN111625954B (en) | Parallel optimization method and system for rainfall landslide model TRIGRS | |
| Pascale et al. | Landslide susceptibility mapping using artificial neural network in the urban area of Senise and San Costantino Albanese (Basilicata, Southern Italy) | |
| CN102354348A (en) | Watershed scale soil moisture remote sensing data assimilation method | |
| Al-shalabi et al. | Manifestation of remote sensing data in modeling urban sprawl using the SLEUTH model and brute force calibration: a case study of Sana’a city, Yemen | |
| Zhang et al. | Double-layer parallelization for hydrological model calibration on HPC systems | |
| Verri et al. | A meteo-hydrological modelling system for the reconstruction of river runoff: the case of the Ofanto river catchment | |
| Carlotto et al. | SW2D-GPU: A two-dimensional shallow water model accelerated by GPGPU | |
| Padulano et al. | Using the present to estimate the future: A simplified approach for the quantification of climate change effects on urban flooding by scenario analysis | |
| Bittinger et al. | Mathematical simulations for better aquifer management | |
| Chen et al. | A computationally efficient hydrologic modeling framework to simulate surface-subsurface hydrological processes at the hillslope scale | |
| Verkaik et al. | GLOBGM v1. 0: a parallel implementation of a 30 arcsec PCR-GLOBWB-MODFLOW global-scale groundwater model | |
| Wei et al. | Enable high-resolution, real-time ensemble simulation and data assimilation of flood inundation using distributed GPU parallelization | |
| Shaad | Development of a distributed surface-subsurface interaction model for river corridor hydrodynamics | |
| Zhang et al. | OpenMP parallelization of a gridded SWAT (SWATG) | |
| Nair et al. | Methods for integrating high-resolution land, climate, and infrastructure scenarios in a hydrologic simulation model | |
| Liu et al. | Parallelization of a distributed ecohydrological model | |
| Bates et al. | LISFLOOD-FP | |
| Zhang | Urban food simulation by coupling a hydrodynamic model with a hydrological model | |
| Nilsson | A dynamic and spatially distributed rainfall runoff model-Developing a model for overland flow in GIS, based on a multiple flow direction algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |